coordax.Field¶

An array with optional named dimensions and associated coordinates.

Examples

>>> import coordax as cx
>>> import jax.numpy as jnp
>>> x = cx.SizedAxis('x', 2)
>>> field = cx.Field(jnp.ones((2, 3, 4)), dims=('x', 'y', None), axes={'x': x})
>>> field
<Field dims=('x', 'y', None) shape=(2, 3, 4) axes={'x': SizedAxis} >
>>> field.data
Array([[[1., 1., 1., 1.],
        [1., 1., 1., 1.],
        [1., 1., 1., 1.]],

       [[1., 1., 1., 1.],
        [1., 1., 1., 1.],
        [1., 1., 1., 1.]]], dtype=float32)
>>> field.dims
('x', 'y', None)
>>> field.axes
{'x': coordax.SizedAxis('x', size=2)}
>>> field.shape
(2, 3, 4)
>>> field.named_shape
{'x': 2, 'y': 3}
>>> field.positional_shape
(4,)
>>> field.named_dims
('x', 'y')
>>> field.named_axes
{'x': 0, 'y': 1}
>>> field.coord_fields
{}
>>> field.coordinate
CartesianProduct(coordinates=(coordax.SizedAxis('x', size=2), coordax.DummyAxis('y', size=3), coordax.DummyAxis(None, size=4)))

Construct a Field.

Parameters:

data – the underlying data array.
dims – optional tuple of dimension names, with the same length as data.ndim. Strings indicate named axes, and may not be repeated. None indicates positional axes. If dims is not provided, all axes are positional.
axes – optional mapping from dimension names to associated coordax.Coordinate objects.

Examples

>>> import coordax as cx
>>> import jax.numpy as jnp
>>> cx.Field(jnp.ones((2, 3)), dims=('x', 'y'))
<Field dims=('x', 'y') shape=(2, 3) axes={} >

See also

coordax.field()

Methods

`__init__`(data[, dims, axes])	Construct a Field.
`astype`(dtype[, copy, device])	Name-vectorized version of array method `astype <numpy.ndarray.astype>`.
`broadcast_like`(other)	Returns a field broadcasted like `other`.
`clip`([min, max])	Name-vectorized version of array method `clip <numpy.ndarray.clip>`.
`conj`()	Name-vectorized version of array method `conj <numpy.ndarray.conj>`.
`conjugate`()	Name-vectorized version of array method `conjugate <numpy.ndarray.conjugate>`.
`from_namedarray`(named_array[, axes])	Creates a Field from a named array.
`from_xarray`(data_array[, coord_types])	Deprecated alias for coordax.from_xarray.
`isel`([indexers])	Returns a new Field with the given indexers applied.
`order_as`(*axis_order)	Returns a field with the axes in the given order.
`round`([decimals, out])	Name-vectorized version of array method `round <numpy.ndarray.round>`.
`sel`([indexers, method])	Returns a new Field with the given selection applied.
`tag`(*names)	Returns a Field with attached coordinates to the positional axes.
`to_xarray`()	Convert this Field to an xarray.DataArray with NumPy array data.
`tree_flatten`()	Flatten this object for JAX pytree operations.
`tree_unflatten`(axes, leaves)	Unflatten this object for JAX pytree operations.
`untag`(*axis_order)	Returns a view of the field with the requested axes made positional.
`unwrap`(*names)	Extracts underlying data from a field without named dimensions.
`view`([dtype, type])	Name-vectorized version of array method `view <numpy.ndarray.view>`.

Attributes

`axes`	This field's coordinates, as a dict of 1d coordinates.
`coord_fields`	A mapping from coordinate field names to their values.
`coordinate`	This field's coordinate, as a single Coordinate object.
`data`	The value of the underlying data array.
`dims`	Named and unnamed dimensions of this array.
`dtype`	The dtype of the field.
`imag`	Name-vectorized version of array method `imag <numpy.ndarray.imag>`.
`named_array`	The value of the underlying named array.
`named_axes`	Mapping from dimension names to axis positions.
`named_dims`	Named dimensions of this array.
`named_shape`	A mapping of axis names to their sizes.
`ndim`	Number of dimensions in the underlying data array.
`positional_shape`	A tuple of axis sizes for any anonymous axes.
`real`	Name-vectorized version of array method `real <numpy.ndarray.real>`.
`shape`	A tuple of axis sizes of the underlying data array.